Google released TPU Sync, an open-source C++ library that moves KV caches and weights across TPU HBM, host DRAM, storage and network. Google reports over 90% lower KV transfer latency on 235B to 480B MoE models. The pos…
Open-weights releases pressure closed pricing and widen access, changing the build-vs-buy math for the whole ecosystem.
Companies and models mentioned in this story — open their pages and live prices
Summaries are aggregated for information only — follow the source link for the full story. Demo entries are illustrative.